You are on page 1of 5

Feesa Ltd Case Study Hydrodynamic Slug Size in Multiphase Flowlines

HYDRODYNAMIC SLUG SIZE IN MULTIPHASE FLOWLINES

Abstract
Figures 1 to 3 show a series of slides from an
Hydrodynamic slugs are understood to be initiated by animation of slug formation available on the Feesa
the instability of waves on the gas-liquid interface in
stratified flow under certain flowing conditions
.
When hydrodynamic slugs are numerous, the pipeline
is said to be operating in the slug flow regime. Th
e
slug flow regime is characterised by complex gas
liquid flows often exhibiting chaotic behaviour.
Commercially available one-dimensional transient
flow simulators rely on correlations to model th
e
complex three-dimensional multiphase phenomena
within a slug. These correlations can be tuned to field
data giving good a posteriori predictions for operating
facilities. However, their ability to make a priori
predictions of slug flow (including slug size an
d
frequency) remains limited. Hence, good design
practice still requires the application of healthy design
margins.
This note considers the subject of hydrodynami
c
slugging and qualitatively addresses the accuracy of
the prediction methods.
Introduction
In many oil and gas developments incorporatin
g
multiphase flowlines, multiphase surges are a major
Flow Assurance concern due to the excessi
ve
demands large changes in oil and gas flow rates place
upon the processing facilities. Multiphase surges
come in three forms:
1. Hydrodynamic Slugs: A property of the stratified
flow regime where slugs are formed due
to
instability of waves at certain flow rates.
2. Terrain Induced Slugs: Caused by accumulation
and periodic purging of liquid in dips along th
e
flowline, particularly at low flow rates.
3. Operationally Induced Surges: Created by forcing
the system from one steady state to another. For
example, during ramp-up or pigging operations.
As these three forms of surges are quite different, they
require different analytical techniques. This note
focuses on hydrodynamic surges, and their prediction.
Slug Formation

Initially the flow is stratified (Figure 1), i.e. the gas is


at the top of the pipe and the liquid is at the bottom.
As gas passes over a wave there is a pressure drop,
then a pressure recovery creating a small forc
Figure 3 Gas Entrainment
e
upward within the wave, analogous to the Bernoulli
effect when gas flows over an aerofoil. Under th
e
correct conditions, this force is sufficient to lift the
wave until it reaches the top of the pipe. This sudden
growth in wave size due to gas flow is triggered by
Once the wave reaches the top of the pipe, it forms
the Kelvin-Helmholtz instability.
into the familiar slug shape (Figure 2), with a nose
Figure 1 Kelvin-Helmholtz Wave Growth
(right hand side) and a tail (left hand side). The slug
is pushed by the gas and so travels at a greater
velocity than the liquid film. Because of this there is
ingress of liquid into the slug nose.
In a Lagrangian frame, the ingress of liquid into the
front of the slug forms a jet which entrains gas
bubbles, as shown in Figure 3. This gas entrainment
Figure 2 Slug Nose Ingress and Tail Shedding
serves to reduce the average liquid holdup in the slug,
interfere with the mechanism of liquid ingress and
increase the turbulence within the slug.
website.
The illustration is presented in a Lagrangian
perspective, following the slug along the pipeline.
Copyright Feesa Limited, 2003
No representation or warranty, express or implied is made by FEESA Ltd as to the completeness, accuracy or fairness of any information contained herein, and FEESA Ltd accepts no
responsibility in relation to such information whether fact, opinion, expectation or otherwise, or for any opinion or conclusion which may be drawn from this document.

Feesa Ltd Case Study Hydrodynamic Slug Size in Multiphase Flowlines

As liquid enters the slug through the front, it als


o
drains under gravity through the tail. The difference
in rates determines the rate at which the slug grows or
decays and ultimately determines the size of the slug.
The rates of liquid ingress and shedding, and t
he
turbulence within the slug, determine whether the slug
will persist or not. These depend on the local flowing
conditions, fluid physical properties and pipelin
e
inclination. In particular, the evolution of slugs is
very sensitive to the pipe inclination and changing the
inclination by less than a degree can be sufficient to
tip the balance causing a flow regime transitio
n.
Thus, peaks and troughs along the pipeline profile of
relatively small amplitude (for example less than
a
pipe diameter) can have a very significant effect.
The possible effect of incorrectly specifying th
e
flowline undulations is illustrated in Figure 4 where
the flow regimes of a simple horizontal topography
(Figure 4a) and a topography with some undulations

(Figure 4b) are compared. The multiphase flow in


Figure 4b switches between the stratified and slug
flow regimes, implying that not only could the slug
sizes differ markedly but the pressure drops could be
very different also. Few pipelines (if any) have
constant inclinations, most undulate following the
natural terrain.
Therefore, when modelling
multiphase flow, it is important to represent these
undulations as faithfully as possible.
Frequency

Slug Growth

Figure 4 The Effect of Pipeline Inclination

a.
Slug Flow

b.

Slug Interaction
Fortunately, observations indicate that slugging
The fact that there are multiple slugs in the pipeline pipelines often obey chaotic attractors and do not
introduces a further complication.
Large flowline normally follow wild excursions in state space. In
systems may have many tens of slugs, generated at design, one is usually interested in the extent of the
different locations, at different times and travelling at attractor (e.g. the maximum holdup in the slug
different velocities. These slugs interact with each catcher). Therefore, simulating an adequate number
other either directly or indirectly, meaning tha of cycles, and sampling with a Poincar section
should be sufficient to establish the picture for design.
t
individual slugs cannot usually be considered i Slug Size Distribution
n
The rich complexity of slug flow combines to produce
isolation.
a range of slug sizes from a given flowline under
A flowline operating in the slug flow regime, often given conditions.
Hence slug size data is often
behaves like a chaotic system exhibiting sensitivity to expressed in frequency-size distribution plots. Such a
the initial conditions. This suggests that in order to distribution is sketched in Figure 5. The plot shows
predict the exact behaviour of a slugging flowline, one that while the mean slug size may be comparatively
must define the initial and boundary condition small, there are less frequent slugs that are many
s
times larger.
perfectly. However, this being said, it is usually not
Figure 5 Typical Slug Size Distribution
necessary
to
predict
the
behaviour
exact
8%
ly.
Stratified Flow

Slug Flow
Slug Flow

Slug Flow

7%

6%
A significant proportion of the frictional pressure drop
in multiphase flow is thought to be due to th 5%
4%
e
turbulent region within the slug. Thus the size of the 3%
turbulent region can have a significant effect on the 2%
frictional pressure losses in a pipeline.
1%
As discussed earlier, an added complication is the
Slug Size
effect of the gas bubbles within the slug that can be 0%
entrained by the liquid ingress into the nose. This
entrainment affects the speed of the slug, the growth
rate and the turbulence, and hence the friction Slug Size Prediction
al
There are two main methods that are typically used
pressure drop of the slug.
for slug size prediction:
A slug is therefore a complex three1. Simple correlations of field data
dimensional
turbulent multiphase phenomenon making it ver
y
Stratified Flow
Stratifiedexactly.
Flow
difficult (arguably impossible)
to model
0

10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

40

42

44

46

48

Copyright Feesa Limited, 2003


No representation or warranty, express or implied is made by FEESA Ltd as to the completeness, accuracy or fairness of any information contained herein, and FEESA Ltd accepts no
responsibility in relation to such information whether fact, opinion, expectation or otherwise, or for any opinion or conclusion which may be drawn from this document.

Feesa Ltd Case Study Hydrodynamic Slug Size in Multiphase Flowlines

2. Transient 1D slug-tracking simulation


Both methods are heavily reliant on good correlations
of field and/or laboratory data, either directly in the
case of the former, or in terms of a clo
sure
relationship for the latter. However there is a limited
amount of slug flow data with sufficient detail t
o
correlate phenomena such as tail shedding, g
as
entrainment, slug interaction, the effect of thre
e
phases and the various other important thr
ee-

dimensional multiphase issues mentioned above.


The available transient slug-tracking simulators use
empirical correlations to compensate for the fact that
their one-dimensional models cannot model these
mechanisms directly. It is unlikely that these
correlations
are
particularly
accurate
and
unfortunately the inherent errors can have a
compound effect on the ultimate slug size. For
example, an error in calculating the balance between
liquid inflow and outflow through a slug of a few
percent, could mean an error in the eventual surge of
several cubic metres per slug. This has obvious
ramifications for the design and operation of slug
catchers.

50

Feedback from operators suggests that slug sizingdiameter vertical pipes and that other less coherent
using transient slug-tracking simulation is less than flow regimes may prevail.
perfect.
Such simulations do, however, give Finally, the advances in Computational Fluid
reasonable predictions when they have been tuned to Dynamics (CFD) and the extension to multiphase
field data. This makes their application as online real flows needs mention since this perhaps offers a longtime simulators of particular value. However, the term solution to multi-dimensional multiphase flows.
engineer charged with designing a new flowline and Many workers have shown how the fundamental
slug catcher system does not this have the benefit of equations of fluid mechanics can be averaged and
field data and must use the tool in a wholly predictive discretised in three-dimensions for multiphase flows
and have produced successful solutions to engineering
manner.
However, as with all of the methods
The simpler (often statistical) correlation methods, are problems.
described,
the
ultimate accuracy depends intrinsically
developed by correlating field slugging results to key
on
the
empirical
relationships that are used to close
parameters such as flowline diameter and flu
the
model,
and
this
is where these advanced methods
id
need
additional
improvement.
Furthermore, for the
physical properties. However, these methods ignore
specific
problem
of
multiphase
flows
in risers, which
the important effect of topography, and can therefore
have
large
length
to
diameter
ratios,
it is difficult to
give misleading results. For example, these methods
see
how
the
application
of
CFD
could
yield practical
can predict slug lengths longer than the flowline
engineering
solutions
without
very
substantial
!
improvements
in
computing
power.
Nonetheless, these correlations have been used on
numerous oil developments and in the absence o Slugcatcher Sizing
f
Slugcatchers should be sized to dampen surges to a
reliable transient predicts do provide another string to level that can be handled by downstream processing
the engineers bow.
equipment. Before dynamic models of the topsides
Slug modelling using the available methods facilities are available, the level of acceptable surging
is unknown and designers are often forced to make
is
stretched even further for large diameter flowlines and assumptions vis--vis surge volumes, such as
risers.
For not only should the correlatio designing for the one in a thousand slug.
However, knowing that the reliability of the
ns
themselves be called into question, but the ve prediction methods is questionable and that there is
always a statistical chance of a larger slug (see Figure
ry
mechanisms on which they are based may also b 5), the possibility of a slug arriving which is greater
than the design surge volume cannot be completely
e
Fortunately, there is often sufficient
incorrect. For example, there is evidence to suggest discounted.
that conventional slug flow does not occur in larg flexibility in the downstream process equipment to
deal with an oversized slug, but even if it does cause a
e
trip this may be acceptable provided that it is
infrequent.
Conclusions
The modelling of hydrodynamic slug flow is in its
infancy and though transient simulation is useful in
calculating the effects of a known slug size they are
still less than perfect at slug size prediction.
Therefore, slug sizing results should always be treated
with caution and slugcatchers should be designed with
an ample design margin.
It is well-known that the potential lost production
arising due to an under-designed slug catcher can be
quite significant. And since the slug catcher usually
represents only a small fraction of the overall
development CAPEX, it is clear therefore that current

best design practice should err on the side of caution.


Copyright Feesa Limited, 2003
No representation or warranty, express or implied is made by FEESA Ltd as to the completeness, accuracy or fairness of any information contained herein, and FEESA Ltd accepts no
responsibility in relation to such information whether fact, opinion, expectation or otherwise, or for any opinion or conclusion which may be drawn from this document.

You might also like